Circular RNA Therapeutics — CMC Emerging Considerations for This Novel RNA Therapeutic Modality

Circular RNA has no 5′ cap. It has no poly(A) tail. The analytical methods validated for linear mRNA drug substance — capping efficiency by LC-MS/MS, poly(A) tail length by CGE, integrity by full-length peak area — are not applicable to circular RNA. Building the quality attributes, analytical methods, and specification limits for this drug substance class means working without FDA-specific guidance, without an approved precedent, and without compendial methods to fall back on.

Circular RNA teams who build their CMC framework rigorously from first principles will define the regulatory standard this modality eventually gets measured against; teams who simply adapt linear mRNA specifications without understanding what is structurally different about a circular molecule are the ones who receive FDA information requests that stall a program for a year or more.

Circularity Purity — Three-Population CGE Analysis, RNase R Treatment, and the Nicked circRNA Immunogenicity Risk

A circRNA drug substance preparation is never a single homogeneous population — it is a mixture of three structurally distinct species that a well-designed CGE method has to resolve independently: the intended circular product, which runs anomalously slowly relative to a linear RNA of equivalent length because its closed topology resists full denaturation; residual linear precursor left over from an incomplete circularization reaction; and nicked or linearized circRNA, generated when the circular backbone breaks after circularization is already complete. That third species is not merely a purity concern — it carries a distinct immunological risk, because a nicked circRNA molecule now presents a free 5′ terminus bearing a triphosphate group, precisely the pattern-associated molecular structure that RIG-I recognizes as foreign RNA, meaning nicked circRNA content is a safety-relevant specification rather than a cosmetic purity number. RNase R treatment, exploiting the enzyme’s inability to process RNA lacking an accessible 5′ terminus, selectively degrades the remaining linear precursor down to roughly 1% or less, but it does nothing to remove nicked circRNA that forms after circularization, meaning that impurity has to be controlled through storage and handling conditions and monitored independently rather than assumed away by the RNase R step. A defensible circularity specification therefore sets individual acceptance criteria for each of the three populations — commonly circular product at 95% or above, with linear precursor and nicked circRNA each held to 5% or below — rather than reporting a single combined CGE peak percentage that cannot distinguish which of the two non-circular species is actually present in a given lot.

Back-Spliced Junction Identity and Full-Length Sequence Confirmation — Why Linear Sequencing Methods Fail for Circular RNA

The back-spliced junction — the point where the 3′ and 5′ ends of the linear precursor are joined by the circularization reaction — is the one sequence feature that exists only in the circular product and nowhere in the linear starting material, which means standard sequencing approaches built around a defined 5′ start point cannot confirm it at all. Detecting the junction requires primers deliberately oriented outward from the predicted junction site, amplifiable only through a rolling-circle reverse transcription event across a circular template, generating a junction-spanning amplicon that can then be Sanger sequenced to confirm the exact nucleotide arrangement flanking the splice point on both sides. Confirming that the rest of the molecule matches the intended sequence — not just the junction itself — requires a complementary whole-sequence method: nanopore-based direct RNA sequencing takes advantage of the circular template’s ability to be traversed repeatedly by a processive RNA helicase, generating concatemeric reads that cover the full sequence multiple times within a single read, and consensus alignment of those reads against the reference sequence can establish full-length identity concordance at or above 99.9% across a sufficiently large read count. A drug substance identity section built entirely on Sanger sequencing from the linear precursor template, without a dedicated BSJ-specific amplification method and a whole-molecule consensus sequencing approach, has confirmed nothing about the one structural feature — the junction — that actually defines the molecule as circular rather than linear.

IRES Translation Efficiency, IRES Immunogenicity, and Group I Intron Fragment Clearance

The internal ribosome entry site driving cap-independent translation of the encoded therapeutic protein is not a passive structural element — its specific sequence identity determines both how efficiently the circRNA translates and how the immune system perceives it, since IRES elements derived from viral genomes carry the sequence and secondary-structure signatures those viruses evolved, and those same signatures can be recognized by innate immune sensors including TLR3, TLR7, and the cytoplasmic dsRNA sensors MDA5 and RIG-I. Different IRES choices produce measurably different translation efficiency relative to conventional cap-dependent translation, meaning the potency assay for a circRNA drug substance has to be IRES-specific: a cell-free translation system using a luciferase reporter, referenced against a characterized standard, with a defensible relative potency specification set well above a minimal functional threshold rather than assumed from the IRES’s published literature performance. The immunogenicity side of this same IRES choice complicates the standard dsRNA quantitation strategy borrowed from linear mRNA CMC, because certain IRES elements carry intrinsic secondary-structure regions that themselves contribute background signal to a J2 antibody-based dsRNA ELISA — meaning a circRNA-specific dsRNA specification has to account for and validate against that IRES-derived background rather than applying a linear-mRNA-derived limit unmodified. Group I intron splicing fragments — small RNA byproducts released during the permuted intron-exon circularization reaction — round out the process-related impurity profile, requiring their own clearance validation, commonly expressed as several orders of magnitude reduction through the downstream purification steps, distinct from and in addition to the circularity purity specification already discussed. A circRNA CMC package that borrows the linear mRNA dsRNA specification unmodified, without accounting for IRES-derived background signal or Group I intron fragment clearance as distinct impurity classes, has applied a framework built for a structurally different molecule.

The XGene Circular RNA Drug Substance CMC Architecture — Manufacturing Process, CQA Specification Design, Validated Analytical Methods, and IRES Immunogenicity Profiling

The XGene Circular RNA Drug Substance CMC Architecture is a structured, first-principles CMC framework built around the recognition that circular RNA’s defining structural features — the absence of a cap or poly(A) tail, the back-spliced junction, and the IRES-dependent translation mechanism — each require dedicated analytical methods that the linear mRNA CMC toolkit does not provide.

1. Manufacturing Process Characterization — Document the permuted intron-exon circularization reaction conditions, the RNase R treatment parameters, and the downstream purification steps as the process controls that directly determine circularity purity outcomes. 2. Three-Population Circularity Specification — Set individual acceptance criteria for circular product, residual linear precursor, and nicked circRNA rather than a single combined purity number. 3. BSJ and Whole-Sequence Identity Methods — Build a divergent-primer RT-PCR and Sanger sequencing method for the junction alongside a nanopore-based whole-sequence consensus method for full-length identity. 4. IRES-Specific Potency and Immunogenicity Profiling — Validate a cell-free translation potency assay specific to the chosen IRES, and account for IRES-derived background signal in the dsRNA specification. 5. Process-Related Impurity Clearance — Validate Group I intron fragment clearance through the purification process as a distinct impurity class from circularity purity.

The output is the circRNA drug substance CMC package built specifically around what makes a circular RNA molecule different from a linear one, rather than a linear mRNA specification framework applied without modification.

FDA’s Guidance for Industry: Chemistry, Manufacturing, and Controls Information for Human Gene Therapy Investigational New Drug Applications (2021) remains the most directly applicable IND CMC guidance for circRNA programs, requiring adaptation of its RNA-based gene therapy framework to circRNA’s specific manufacturing and analytical challenges. Early circRNA IND filings from companies including Orna Therapeutics and Laronde represent the field’s first publicly disclosed regulatory engagement with FDA on this modality, establishing that circRNA is reviewable as a novel RNA therapeutic under the existing gene therapy guidance framework even without modality-specific guidance. The permuted intron-exon circularization methodology that underlies most current circRNA manufacturing platforms traces to foundational academic work on Group I intron-based splicing published in the late 2010s.

For your circular RNA IND CMC package, can you confirm today that your circularity purity specification uses a three-population CGE method capable of independently distinguishing circular product, nicked circRNA, and linear precursor, and that your BSJ sequence identity method uses divergently oriented RT-PCR primers rather than standard linear sequencing approaches that cannot detect the junction at all?